codex-cli-portable-setup-kit

[!TIP] Community Link & Acknowledgements: This project is recognized by and affiliated with the LINUX DO Community (https://linux.do) open-source initiative.

This is a community-maintained Codex configuration and installer. It is not Codex itself and is not an official OpenAI product.

Setup Kit Family

Highlights

• Action over advice. Codex is told to use available tools and change the real target instead of replying with a tutorial you still have to execute.
• Deterministic Reverse Engineering (re-toolkit). Integrated pure-Python subsystem for PE32/PE32+ analysis, x86/x64 disassembly, AOB pattern scanning, CPU micro-emulation, Protobuf/TLV wireformat dissection, and Frida dynamic hook generation.
• Dual-Stage Decoupled Pipeline. Eliminates token degeneration and robotic placeholder artifacts (ROLE_A, ROLE_B) by separating structural planning from high-fidelity rendering.
• Zero-Friction Elevated Execution (TURBO Mode). Pre-approves runtime tools, configures Windows elevated sandboxing, enables non-interactive autonomy (approval_policy = "never"), and marks workspaces as fully trusted.
• Unified Multi-Agent Compatibility. Ships native out-of-the-box instructions for OpenAI Codex CLI (AGENTS.md), OpenCode (OPENCODE.md, opencode.json), Anthropic Claude Code (CLAUDE.md), Cursor (.cursorrules), Windsurf (.windsurfrules), and ChatGPT App (docs/CHATGPT_APP_PRESET.md).
• One-Click Universal Adaptor (ADAPT_ALL.cmd). Automatically configures and syncs all GPT-based agent frameworks and IDEs in your workspace simultaneously.
• Transactional Verification & Rollback. SHA-256 manifest checks, transactional backups, receipts, and one-click rollback (rollback.ps1 / ROLLBACK.sh).

What this profile changes

This project is not mainly a migration utility. Portability is the delivery mechanism. The product is a tuned operating style for Codex. Do the work, then report the result.

Quick start

🪟 Windows 10 & 11 (Primary Platform — 1-Click Launch)

Extract the release ZIP and double-click:

TURBO.cmd

This automatically validates the workspace, configures elevated execution policies, inherits your native model settings, and launches Codex ready for work.

To adapt all agent platforms (Codex, OpenCode, Cursor, Windsurf, Claude Code) in 1 second:

ADAPT_ALL.cmd

🍏 macOS & 🐧 Linux / WSL (Cross-Platform Execution)

bash ADAPT_ALL.sh
# Or launch directly:
bash TURBO.sh

Manual Transactional Installation

Set-ExecutionPolicy -Scope Process Bypass
.\install.ps1 -ProjectRoot 'C:\path\to\your-workspace' -KeysmithPreset unrestricted -SkipPlugins

Verify Package Integrity

.\verify.ps1
# Or verify installed workspace:
.\verify.ps1 -Installed -ProjectRoot 'C:\path\to\your-workspace'

re-toolkit Subsystem

The built-in re-toolkit (payload/project/.agents/tools/re-toolkit/) provides deterministic ground truth for systems engineering and security analysis:

re-toolkit/
├── pe_parser.py          # PE32/PE32+ Header, Section Table, Imports/Exports, RVA translation
├── disasm.py             # x86/x64 Disassembler, AOB wildcard scanner, binary patch generator
├── emulator.py           # Micro-architecture CPU register and stack sandbox emulator
├── protocol_parser.py    # Schema-less Protobuf wire format and TLV frame dissectors
├── frida_bridge.py       # Frida Interceptor hook & anti-debug bypass script generator
├── pipeline/             # Dual-stage decoupled planning & rendering engine
└── cli.py                # Unified CLI entry point

CLI Command Reference

# 1. Automatic triage and format discovery
python .agents/tools/re-toolkit/cli.py auto target.dll --json

# 2. Inspect PE headers, imported DLLs, and exported symbols
python .agents/tools/re-toolkit/cli.py parse-pe sample.exe --json

# 3. Fast AOB hex pattern scanner with wildcards
python .agents/tools/re-toolkit/cli.py pattern-scan target.bin --pattern "48 89 ?? 24 ?? 55" --json

# 4. Extract ASCII & UTF-16LE strings with offsets
python .agents/tools/re-toolkit/cli.py strings sample.exe --min-len 5 --json

# 5. Disassemble raw opcodes or binary sections
python .agents/tools/re-toolkit/cli.py disasm 90505831C0C3 --arch x86_64

# 6. Dissect undocumented Protobuf wireformat payloads
python .agents/tools/re-toolkit/cli.py decode-protobuf 089601120474657374

# 7. Emulate CPU registers and stack execution in isolated memory sandbox
python .agents/tools/re-toolkit/cli.py emulate --code B82A000000505BC3

# 8. Generate Frida dynamic API interceptor and anti-debug scripts
python .agents/tools/re-toolkit/cli.py gen-hook --symbol CryptHashData --module advapi32.dll --output hook.js

# 9. Calculate binary diff patches
python .agents/tools/re-toolkit/cli.py diff-patch --orig orig.bin --patched mod.bin --json

# 10. Format aligned hexadecimal dumps
python .agents/tools/re-toolkit/cli.py hexdump target.bin --offset 0x100 --length 128

# 11. Speculative dual-model shadow probe with instant refusal abort & failover
python .agents/tools/re-toolkit/cli.py spec-probe "Reverse engineer target module and extract memory offsets" --target game.dll --json

# 12. Export universal presets for ChatGPT App, OpenCode, Cursor, and Windsurf
python .agents/tools/re-toolkit/cli.py export-preset --target all

Dual-Stage Decoupled Pipeline

Traditional adversarial constraint checklists often overload LLMs, resulting in attention collapse, generic placeholder tokens (ROLE_A, ROLE_B), and truncated outputs.

The Dual-Stage Decoupled Pipeline resolves this through structural separation:

flowchart LR
    A[User Request] --> B[Stage 1: Structural Planner]
    B -->|SceneBlueprint & Plot Beats| C[Stage 2: High-Fidelity Renderer]
    C -->|Natural, Uninhibited Output| D[Delivered Result]

python .agents/tools/re-toolkit/cli.py pipeline "Your complex scenario or system architecture prompt"

Why not copy .codex manually?

A one-line prompt or copied config.toml can change tone, but the full behavior also depends on instructions, project rules, skills, hooks, feature flags, and continuity tools. This kit installs that tested stack together and gives you a way back.